基于盲辨识模型和极限学习机的下肢假肢路况识别

摘要
图/表
参考文献(13)
相关文章 (8)

全文: PDF (1151 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要针对大腿截肢者穿戴智能假肢路况识别准确率低的问题，提出了一种盲辨识理论和极限学习机相结合的路况识别方法。选择肌电信号(EMG)作为路况识别信息源，提取肌电信号的盲辨识模型系数作为信号特征，为了能够充分反映路况特征,比较了不同特征值，分析了选取盲辨识模型系数作为路况识别特征值的合理性。为了克服极限学习机(ELM)分类器随机产生的输入权值只有少部分是比较优越的缺点，利用烟花极限学习机（FA-ELM）对平地行走、上楼、下楼、上坡、下坡、跑步6种路况进行分类。与ELM算法、BP神经网络进行了对比，结果表明, 采用盲辨识模型和烟花算法进化极限学习机将6种路况下平均识别率提高到93.18%，优于ELM和BP神经网络等方法。


	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	刘磊1
	杨鹏2
	刘作军2
	宋寅卯1
	吴青娥2

关键词 ：肌电信号, 盲辨识模型, 极限学习机, 路况识别, 智能假肢

Abstract：In order to solve the problem of lower road condition recognition rate of leg amputees with intelligent prostheses, a road condition recognition method based on blind recognition theory combined with extreme learning machine was proposed.Firstly, surface electro-myographic (sEMG）signal was selected as the road condition identification information source, and blind identification model coefficients of sEMG signal were extracted as the signal features.In order to fully reflect the road condition features, different feature values were compared to analyze the reasonability of choosing blind identification model coefficients as road condition recognition features.In order to overcome the disadvantage of only a few input weights generated randomly by ELM classifier being superior, a firework ELM was used to classify 6 road conditions including walking on flat ground, going upstairs, going downstairs, going uphill, going downhill and running.Finally, the results using the proposed method were compared with those using ELM algorithm and BP neural network, respectively.It was shown that using the blind identification model and the ELM with firework evolution algorithm, the average recognition rate of 6 road conditions is increased to 92.51%, it is superior to those using ELM algorithm and BP neural network, respectively.

Key words： surface electro-myographic (sEMG) signal blind identification model extreme learning machine (ELM) road condition recognition intelligent prosthesis

收稿日期: 2019-02-12 出版日期: 2019-11-28

引用本文:

刘磊1，杨鹏2，刘作军2，宋寅卯1，吴青娥2. 基于盲辨识模型和极限学习机的下肢假肢路况识别[J]. 振动与冲击, 2019, 38(23): 178-185.
LIU Lei1,YANG Peng2,LIU Zuojun2,SONG Yinmao1, WU Qinge2. A road condition recognition method for leg amputees with intelligent prostheses based on blind identification model and extreme learning machine. JOURNAL OF VIBRATION AND SHOCK, 2019, 38(23): 178-185.

链接本文:

http://jvs.sjtu.edu.cn/CN/ 或 http://jvs.sjtu.edu.cn/CN/Y2019/V38/I23/178

[1]David Q, Anne E M, Robert D. G. Toward Unified Control of a Powered Prosthetic Leg: A Simulation Study[J]. IEEE Trans. on control systems technology, 2018, 26(1): 305 -312.
[2]金德闻, 张济川, 郝智秀。康复工程中的生物机械学研究[J].中国康复医学杂志，2010，25(1): 61-63.
JIN De-wen, ZHANG Ji-chuan, HAO Zhi-xiu. Biomechanical research in rehabilitation engineering [J]. Chinese Journal of Rehabilitation Medicine, 2010, 25(1): 61-63.
[3]Chen Bao-jun , Wang Xue-gang , Huang Yan , Wei Kun-lin , Wang Qi-ning. A foot-wearable interface for locomotion mode recognition based on discrete contact force distribution[J]. Mechatronics, 2015, 32(1): 12 -21.
[4]Huang S, Ferris D P. Muscle activation patterns during walking from transtibial amputees recorded within the residual limb-prostheticinterface[J]. Journal of NeuroEngineering and Rehabilitation, 2012, 9(1): 1 -16.
[5]佟丽娜，侯增广，彭亮。基于多路sEMG 时序分析的人体运动模式识别方法[J]。自动化学报，2014，40(5): 810- 820.
TONG Li-na, HOU Zeng-guang, PENG Liang. Multi-channel sEMG Time Series Analysis Based Human MotionRecognition Method [J]. Acta automatica sinica, 2014, 40(5): 810-820.
[6]HUANG S, HUANG H. Voluntary control of residual antagonistic muscles in transtibial amputees:feedforward ballistic contractions and implications for direct neural control of powered lower limb prostheses[J].IEEE Transaction on neural systems and rehabilitation engineering,2018,26(4): 894-903.
[7]李影．基于sEMG的步态运动模式细分及其识别方法 [D]．杭州：杭州电子科技大学，2018.
LI Ying. Gait Pattern Segmentation and Gait Recognition Based on sEMG Signals [D]. Hang zhou: Jilin University, 2018.
[8] 尚小晶．肌电-运动关系建模与运动模式识别研究 [D]．长春：吉林大学，2012.
SHANG Xiao-jing. On the sEMG-Movement Relationship Modeling and Movement Pattern Recognition [D]. Chang chun: Jilin University, 2012.
[9]Ding F, Chen T. Identification of Hammerstein nonlinear AR-MAX systems [J]. Automatica, 2005, 41(9):1479-1489.
[10]秦斌，易怀洋，王欣．基于极限学习机的风电机组叶根载荷辨识建模[J].振动与冲击，2018,37(4): 257-262.
QIN Bing, YI Huai-yang, WANG Xin. A model of wind turbine blade root loads based on extreme learning machine[J]. Journal of Vibration and Shock, 2018,37(4): 257-262.
[11]李一博，祝慧宇，张玉祥，等. 基于极限学习机和b值法的罐底声发射检测腐蚀信号识别方法[J].振动与冲击，2015,34(11): 35-40.
LI Yi-bo, ZHU Hui-yu, ZHANG Yu-xiang,etal.Tank bottom corrosion signals recognition with acoustic emission testing based on ELM and b-value method[J]. Journal of Vibration and Shock, 2015,34(11): 35-40.
[ [12]郭欣，王蕾，宣伯凯，等．基于有监督 Kohonen神经网络的步态识别[J]. 自动化学报，2017，43(3)：430-438.
GUO Xin, WANG Lei, XUAN Bo-kai, etal.Gait Recognition Based on Supervised Kohonen Neural Network [J]. Acta automatica sinica, 2017, 43(3): 430-438.
[13]吴剑锋．基于肌电信号的人体下肢运动信息获取技术研究 [D]．杭州：浙江大学，2008.
WU Jian-feng. Research on Human Lower-Limb Motion Information Acquisition Technology Based on EMG [J]. Hang zhou: Zhejiang University, 2008.